Myb-dependent Regulation of Thrombospondin 2 Expression

The nuclear transcription factor c-Myb, which is highly expressed in hematopoietic cells, has been shown to be functional in NIH 3T3 cells: cells that do not possess detectable levels of c-Myb. To identify endogenous target genes of c-Myb in fibroblasts, RNA isolated from NIH 3T3 cells stably transfected with a full-length or a dominant negative c-myb construct (GREMyb and GREMEn, respectively) was subjected to differential display analysis. 5′-Rapid amplification of cDNA ends of a selected band, sequencing, and a nucleotide homology search led to the identification of thrombospondin 2 (TSP 2) as the gene product repressed in GREMyb and induced in GREMEn cells. The pattern of TSP 2 expression during the cell cycle was consistent with c-myb-dependent regulation. The possibility that the identified transcript was TSP 1, a homologous product known to be repressed by v-Src, c-Jun, and v-Myc, was ruled out by using a TSP 2-specific DNA probe and by showing a distinct pattern of regulation of TSP 1 and TSP 2 expression. Nuclear run-on and TSP 2 promoter-reporter (chloramphenicol acetyltransferase) assays showed similar transcriptional levels in GREMyb and NIH 3T3 cells. However, mRNA stability studies showed a much shorter TSP 2 mRNA half-life in GREMyb compared with wild type NIH 3T3 cells, suggesting that c-myb affects TSP 2 expression via a post-transcriptional mechanism. The implications of a protooncogene-mediated suppression of TSP expression are discussed.

G 1 /S progression is not known, the protooncogene has been shown to control intracellular ionized calcium levels by modulating expression of the plasma membrane Ca 2ϩ -ATPase pump (16). In addition, c-myb may be involved in regulation of transcription of several cell cycle enzymes including DNA polymerase ␣, cdc2, and proliferating cell nuclear antigen (17)(18)(19).
The c-Myb protein contains three distinct functional domains: a DNA binding domain, a transactivation domain, and a negative regulatory domain (20). Although a c-myb-dependent transcriptional transactivation that is independent of the DNA binding domain has been reported (21), all the remaining c-myb target genes that have been identified to date contain the sequence PyAACG/TG that is recognized by the c-Myb DNA binding domain (22). The number of promoters that have been found to contain the c-Myb binding site has risen to more than 10 because of the expected high redundant occurrence of a hexanucleotide sequence (ϳ10 6 ) in the mammalian genome (1). Thus, more studies are needed to show that identified genes represent true endogenous target genes. This consideration is especially important given the broad spectrum of myb-associated cellular events. One such potential target gene is mim-1, the product of which is a protein of unknown function specifically expressed in promyelocytes (23).
Although considered to be a transcriptional activator, c-myb has been shown to repress the transcription of constructs containing the c-erbB-2 promoter (24) or the c-Myb consensus binding site itself (25). In addition, recent studies of c-myb-dependent regulation of the plasma membrane Ca 2ϩ -ATPase pump suggest oncogene-dependent down-regulation of its expression.
In order to identify myb-responsive genes involved in the regulation of cell growth, we studied c-myb-dependent gene expression in an NIH 3T3 cell line possessing myb-dependent cell cycle machinery but demonstrating no detectable levels of endogenous c-myb expression (26). The cells were stably transfected with a full-length murine c-myb cDNA sequence as well as a chimeric construct composed of a c-myb DNA binding domain linked to the Drosophila engrailed transcription suppressor that was shown to act in a dominant-negative manner in these cells (26) and subjected to differential display analysis. Comparison of these three cell lines then allowed for identification of either c-myb-induced or -suppressed genes.
On the basis of these studies, we found that the expression of the extracellular matrix protein thrombospondin (TSP) 1 2, re-cently shown to inhibit neovascularization induced in the rat cornea (27) and to exhibit increased vascular density in many tissues of TSP 2 knockout mice (28), demonstrated c-myb-dependent regulation of expression. Interestingly, this c-myb-dependent control of TSP 2 expression was found to occur at the post-transcriptional level. This finding of a protooncogene-dependent suppression of one of the principal angiogenesis inhibitor genes highlights an unexpected function of c-myb and suggests an interesting mechanism of tumor growth.
The expression constructs pGRE-Myb and pGRE-MEn and the derived cell lines GREMyb and GREMEn selected for the expression of the full-length c-myb and the dominant negative MEn, respectively, have been described previously (26). The murine TSP 2 CAT plasmid, containing a 1208-base pair genomic fragment (nucleotides 92 to -1116 according to the numbering by Bornstein et al. (29)), was generated by PCR amplification of genomic DNA from NIH 3T3 cells, using a 5Ј primer: ATG GAT CCA ACA CAA TGG CAG G starting at nucleotide -1116, and a 3Ј primer: ACG CCA GTA CTC TGT CTG TC starting at nucleotide 92, and cloned in the vector pCAT-basic (Promega).
Transient Transfections and CAT Assays-Transient transfections were carried out using the calcium phosphate precipitation method (30,31). For each round of transfections, cells were plated in six-well dishes (Falcon) and incubated with supercoiled plasmid DNA in the presence of calcium phosphate precipitate for 4 -5 h and then washed three times after a brief glycerol shock. In all experiments, transient transfection mixtures included 0.5 g of pCMV-␤-gal to assess transfection efficiency.
For determination of CAT activity, cells were harvested and lysed, and the cell extract assayed for ␤-gal and CAT activity using the Promega assay kit protocol. CAT activity was determined using a liquid scintillation counting assay.
RNA Preparation and Northern Blot Analysis-Total RNA from NIH 3T3 and stable transfectant clones was prepared by centrifugation through a CsCl gradient or by using TriReagent solution (Sigma), while poly(A) ϩ RNA was prepared using total RNA isolated from NIH 3T3 cells and an mRNA purification kit (QIAGEN). For Northern blot analysis, RNA samples (5-15 g/lane) were fractionated by electrophoresis on a 1.3% agarose formaldehyde gel and transferred to GeneScreen Plus membrane (NEN Life Science Products). Hybridization using 32 P-labeled probes was performed at 68°C for 1.5-3 h in Quickhyb solution (Stratagene). Following hybridization, filters were washed twice at room temperature in 2ϫ SSC, 0.1% SDS for 5-10 min each and at 55 to 68°C in 0.1ϫ SSC, 0.1% SDS for 15-30 min and subjected to autoradiography. All cDNA probes were prepared by random primer labeling, followed by purification using a Sephadex G-50 spin-column (Boehringer Manheim).
The probes used to detect TSP 1 and 2 mRNAs corresponded to the 1.5-kb-long 5Ј end region of murine TSP 1 cDNA (a fragment extending to the EcoRI site at nucleotide 1404; Ref. 32) and a 500-base pair PstI/PstI fragment spanning exons 4 -7 of the murine TSP 2 cDNA (27), respectively. These probes were a kind gift of Dr. M. L. Iruela-Arispe (Beth Israel Deaconess Medical Center, Boston, MA). In the initial studies, a 5Ј-RACE product based on differential display analysis was used as a probe. The 36B4 probe used to detect the acidic ribosomal phosphoprotein PO mRNA (33) was a kind gift of Dr. Kevin Claffey (Beth Israel Deaconess Medical Center, Boston, MA).
Differential Display Analysis-Differential display analysis was carried out following the Gene Hunter kit protocol as described (34). Total RNA was isolated from near-confluent cultures subjected to 24 -36 h of serum starvation (0.5% FBS, 0.1% BSA-DMEM), treated with RNasefree DNase, and reverse-transcribed in a 10-l reaction mixture with 5-10 units of avian myeloblastosis virus super-reverse transcriptase (Molecular Genetic Resources), 20 M dNTP, and 1.0 M of one of the four anchoring primers T12MN for 60 min at 37°C. Two-l aliquots of the reverse-transcribed sample were PCR-amplified (using Ampli-Taq) in the presence of 2.0 M dNTP and [␣-35 S]dATP, 0.2 M 5Ј arbitrary primer, and 1 M T12MN, the same 3Ј primer that was used for reverse transcription. The parameters for PCR amplification were 94°C for 30 s, 40°C for 2 min, 72°C for 30 s with 40 cycles, and a final 5-min elongation time at 72°C.
Amplified products were fractionated on a 6% sequencing polyacrylamide gel. Following fractionation, the gel was placed on Whatman no. 3MM backing paper, dried, and exposed to XAR-5 film. The cDNA bands of interest were excised by cutting through the film with a razor blade and eluted by soaking and boiling (10 and 15 min, respectively) in 100 l of distilled H 2 O. The DNA in the supernatant was then precipitated with ethanol and resuspended in 10 l of distilled H 2 O. Four l of the cDNA template were re-amplified using the same primer set and PCR conditions, except the dNTP concentration was 20 M with no isotope added. The re-amplified samples were blunt-ended by treatment with T4 DNA polymerase and cloned in the EcoRV site of pBluescript. Nucleotide sequence was determined by automated sequencing and compared against the GenBank data base using the Blast program.
To extend the sequence of the obtained differential display band, designated band 13, a specific oligonucleotide primer (GTGGCAAT-GTCTCGTGATGTGG) was used to reverse transcribe poly(A) ϩ RNA isolated from NIH 3T3 cells. The first strand cDNA product was PCRamplified in the presence of the same 3Ј primer used for reverse transcription and a 5Ј primer based on the Kozak consensus sequence (GCCACCATGG). The PCR end product was cloned in pBluescript and sequenced.
Western Blotting-Cells grown to confluence were counted and plated on 100-mm dishes (Falcon, 4 ϫ 10 6 cells/dish) in 10% FBS-DMEM, for approximately 4 h to allow attachment. The medium was replaced with 0.5% FBS-DMEM supplemented with 0.1% BSA and the cells were starved for 20 -24 h. Following starvation, the cells were rinsed twice with serum-free medium and incubated in the same medium. Six hours later, the cells were detached using PBS containing 0.05 mM EDTA, centrifuged, and lysed in radioimmune precipitation buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM EGTA, 5 mM EDTA, 20 mM NaF, 20 mM sodium pyrophosphate, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride). The lysate was centrifuged at 5000 rpm for 15 min. Protein concentration was determined by absorbance at 595 nm using protein assay reagent (Bio-Rad), and 15 g of protein were separated on 6% polyacrylamide gel. After transfer to nitrocellulose membrane (Amersham Pharmacia Biotech) overnight at 24 V, the membrane was rinsed twice (5 min each) in TBS, and blocked for 1 h with 0.2% nonfat milk in TBS. The membrane was then incubated with rabbit anti-mouse TSP 2 antibody (a kind gift of Dr. P. Bornstein, University of Washington, Seattle, WA) in 0.2% nonfat milk in TBS containing 0.1% Tween 20. The membrane was washed and then incubated with alkaline phosphatase-conjugated goat anti-rabbit IgG. The blot was developed using an Immun-Star chemiluminescent protein detection system (Bio-Rad).
Thrombospondin 1 and 2 mRNA Stability-Cells grown to confluence were counted and plated on 100-mm dishes (Falcon, 4 ϫ 10 6 cells/dish) in 10% FBS-DMEM, for approximately 4 h to allow attachment. The medium was replaced with 0.5% FBS-DMEM supplemented with 0.1% BSA and incubated for 20 h, after which actinomycin D, an inhibitor of transcription, was added (25 g/ml). RNA was isolated at the indicated time points (see "Results") and stored at -70°C until all samples were collected. Northern blot analysis was performed using TSP-1-and TSP-2-specific probes. The signal was quantified by Phos-phorImager analysis, standardized using the acidic ribosomal phosphoprotein PO mRNA 36B4 as an internal control, and plotted as the percent of RNA remaining versus the time after addition of actinomycin D. The slope of the linear portion of the decay curve was calculated and the mRNA half-life determined using the formula t1 ⁄2 ϭ ln2/mRNA decay rate constant (35).

RESULTS
Differential Display Analysis of c-myb-dependent Gene Expression-In order to identify genes regulated by c-myb expression, we employed differential display analysis of myb-dependent transcription in growth-arrested NIH 3T3 fibroblasts as well as in 3T3-derived clonal cell lines GREMyb and GREMEn. c-myb-overexpressing cells (GREMyb) were generated by stable expression of an inducible c-myb construct pGRE-Myb containing full-length murine c-myb cDNA sequence under the control of a glucocorticoid-sensitive promoter GRE, while the GREMEn cell line was generated by stable expression of a dominant-negative construct, pGRE-MEn, containing Drosophila engrailed gene linked to c-myb DNA binding sequence. The generation and characterization of these cell lines and physiologic effects of c-myb and c-myb dominant-negative construct expression in these cells have been described previously (26,36). The choice of 3T3 cells for these experiments was dictated by low (but physiologically significant) levels of Myb activity and by demonstrated sensitivity to manipulations of c-myb expression (26).
To minimize any impact of growth-dependent gene expression on the results of differential display, the cell lines used in these experiments were growth-arrested for 24 h prior to RNA isolation. Five differentially displayed bands were reproducibly observed on multiple runs of total cytoplasmic RNA using a standard protocol (34). One of the bands demonstrated increased expression in a myb-overexpressor cell line GREMyb, while expression of four other bands was strongly inhibited in GREMyb cells.
One of these genes was identified and fully characterized. The band was isolated from the differential display gel and cloned into the plasmid pBluescript. Northern analysis using a cDNA probe derived from four different plasmid clones identified mRNA species of identical size demonstrating the same pattern of expression as seen originally on differential display (not shown), suggesting that a single gene transcript was present in the cloned differentially expressed band.
In order to extend the sequence of the cloned fragment, we applied 5Ј-RACE on mRNA isolated from NIH 3T3 cells using a 20-mer oligonucleotide primer complementary to a cloned differential display band sequence as a 3Ј primer, and an oligonucleotide primer based on the Kozak consensus sequence as a 5Ј primer. The procedure yielded a 1.8-kb product that was subjected to further analysis. To verify that the expression pattern of the cloned transcript was the same as that of the original differential display band, we carried out Northern analysis on total RNA isolated from cells synchronized in the G 0 phase of the cell cycle by prolonged serum starvation. The probe hybridized to a 6-kb transcript that was down-regulated in c-myb-overexpressing cells compared with wild type NIH 3T3 cells or 3T3 cells transfected with the dominant negative c-myb construct pGRE-MEn (Fig. 1A, TSP 2 panel; compare  lanes 3 and 4 with lanes 1 and 2 and 5 and 6).
Partial sequence analysis of the cloned product demonstrated a 100% homology to nucleotides 3337-3551 of the mouse TSP 2 cDNA sequence and a 77% homology to murine TSP 1 (nucleotides 3321-3537 of the sequence). Thus, the cloned product appeared to be the mouse TSP 2 cDNA. To   FIG. 1. c-Myb effects on TSP gene expression. Northern blot analysis of total RNA (15 g) isolated from the wild type NIH 3T3 (lanes 1 and 2), GREMyb (lanes 3 and 4) and GREMEn cells ( lanes  5 and 6). A, serum-starved cells in the absence (Ϫ) and presence (ϩ) of 0.5 M dexamethasone (Dex). The blot was probed with a random-primed 5Ј-RACE extension product obtained by using a 3Ј end oligonucleotide primer based on the differential display analysis (GTG-GCAATGTCTCGTGATGTGG) and a 5Ј end oligonucleotide primer based on the Kozak consensus sequence (GCCAC-CATGG) as well as thrombospondin 1-specific and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probes (55). B, quiescent cells before (Ϫ) or following (ϩ) stimulation with 10% FBS (ϩ) for 16 h. The blot was probed with thrombospondin 1 and thrombospondin 2-specific DNA probes (see "Experimental Procedures"). confirm this identification further, Northern analysis with a TSP 2 cDNA probe was carried out demonstrating hybridization to the same band as detected by the cloned product (data not shown).
Considering earlier studies on the inhibition of TSP 1 expression by v-Src, c-Jun, and v-Myc (37-39) and the extent of sequence homology between the cloned 5Ј-RACE product and TSP 1 (77%), the same blot was hybridized with a TSP 1-specific probe. As with TSP 2, we found that TSP 1 mRNA levels were down-regulated in c-myb-overexpressing cells compared with NIH 3T3 and GREMEn cells (Fig. 1A, TSP 1 3 and 4 with lanes 1 and 2 and 5 and 6).
c-myb-dependent Regulation of Thrombospondin 2 Gene Expression-To verify that the two signals represented two distinct transcripts (i.e. TSP 1 and TSP 2), and to study the potential link between c-myb and TSP gene expression, we analyzed the appearance of the gene transcripts for TSPs 1 and 2 in serum-stimulated NIH 3T3 cells, and in the clones GRE-Myb and GREMEn, using TSP 1-and TSP 2-specific cDNA probes. The pattern of expression of TSP 1 and TSP 2 following serum stimulation of quiescent NIH 3T3 and GREMyb cells shows a clear difference in the magnitude of induction (Fig.  1B). Exposure of GREMyb cells to serum leads to induction of TSP 1 mRNA to levels equivalent to that seen in serum-stimulated control cells. On the other hand, expression of TSP 2 in GREMyb cells following exposure to serum remained lower than expression in control 3T3 cells. In both serum-starved and serum-stimulated cells, TSP 2 mRNA levels were 5-fold less in GREMyb than in NIH 3T3 cells. This suggests that serum stimulation can acutely (but not chronically) override the suppression of TSP 1 gene expression by c-Myb but not that of TSP 2.
Studies of the differential expression of both TSP genes described above have relied on the ability of the two chosen cDNA probes to distinguish the two transcripts. To directly address the specificity of the selected probes for TSP 1 and 2 gene expression, we analyzed the effect of cell density on TSP 1 and TSP 2 mRNA levels. To this end, wild type NIH 3T3 cells were grown to confluence, harvested, and seeded at different concentrations. Sixteen hours later, the cells were harvested for RNA extraction and analyzed for TSP 1 and TSP 2 expression with cDNA probes specific for each gene. In agreement with previous studies (40), TSP 1 expression varies inversely with cell density, while TSP 2 expression varies directly with cell density (Fig. 2). These results show a basic difference in the regulation of TSP 1 and TSP 2 expression and confirm the specific detection of TSP 1 and TSP 2 gene products by the selected cDNA probes.
To determine whether the down-regulation of TSP 2 transcripts could be extended to the protein level, we performed Western blot analysis using a rabbit anti-mouse TSP 2 antibody. A band of the expected size (200 kDa) was detected. Consistent with the Northern blot analysis, less TSP 2 was expressed by GREMyb cells than by NIH 3T3 or GREMEn cells (Fig. 3).
Because c-myb is expressed in a cell cycle-dependent manner, suppression of genes dependent on c-myb regulation should demonstrate an inverse temporal expression pattern. Therefore, we studied cell cycle dependence of TSP 1 and 2 gene expression following serum stimulation of synchronized NIH 3T3 cells. In agreement with prior observations (32), both genes exhibited an early increase in expression following serum stimulation, with TSP 1 expression levels peaking earlier than those of TSP 2 (Fig. 4). However, although TSP 1 levels, following the early serum-sensitive peak, remained relatively unchanged throughout the cell cycle, TSP 2 levels demonstrated a marked reduction in expression between 8 and 16 h (Fig. 4), a time period corresponding to a rise in c-myb expression (12). We conclude from these observations that the pattern of TSP 2 gene expression is consistent with c-Myb-dependent regulation.
c-Myb Expression and Transcriptional Regulation of Throm-

FIG. 2. Differential regulation of thrombospondin 1 and 2 gene expression by cell density.
Northern analysis of total RNA (5 g/lane) isolated from NIH 3T3 cells that were grown to confluence, enumerated, and seeded in 10-cm dishes at the density indicated. The cells were cultured, after seeding, in 10% FBS for 16 h prior to harvesting of RNA. The blot was probed with thrombospondin 1 and thrombospondin 2-specific cDNA probes. An 18 S rRNA band stained with ethidium bromide is also shown. bospondin Expression-Because c-Myb is a DNA-binding transcriptional factor that has been shown to directly affect expression of several different promoters (17)(18)(19), we tested c-Myb's ability to affect TSP 2 gene expression. Transient transfection studies of the murine TSP 2 promoter linked to a CAT reporter construct showed no difference in CAT activity in GREMyb compared with wild type 3T3 cells (Fig. 5).
To examine whether the results seen in transient CAT assays paralleled the pattern of regulation of endogenous gene expression, we performed nuclear run-on assays. Radiolabeled transcripts obtained from nuclei of NIH 3T3 and GREMyb cells cultured in 0.5% FBS, 0.1% BSA for 24 h were hybridized to TSP 2 and ␤-actin probes (Fig. 6). PhosphorImager analysis of the bands showed that, compared with that in NIH 3T3 cells, transcription of TSP 2 was somewhat (1.2-fold) increased in GREMyb cells. This suggested that suppression of TSP 2 expression in GREMyb cells was mainly effected post-transcriptionally.
Thrombospondin 2 mRNA Stability Studies-To investigate the role of mRNA stability in myb-dependent regulation of TSP 2 expression, levels of TSP 2 mRNA were assessed by Northern analysis of RNA isolated from GREMyb and wild type NIH 3T3 cells cultured in the presence of actinomycin D (Fig. 7A). We found a significant reduction in the TSP 2 message half-life in c-myb-overexpressing cells (5 h in GREMyb versus 10 h in 3T3 cells, Fig. 7B, left panel). In contrast, there was no significant difference in the rate of TSP 1 message degradation between NIH 3T3 and GREMyb cells (Fig. 7B, right panel). These experiments suggest, therefore, that decreased TSP 2 mRNA stability in the presence of c-myb expression may be the primary mechanism for myb-dependent suppression of TSP 2 expression. DISCUSSION c-Myb is considered to be a transcriptional effector molecule, which plays an important role in regulation of cell growth and differentiation. Although most published studies to date suggest that c-myb acts as a transcriptional transactivator, previous investigations have suggested that the protooncogene may act as a repressor (20,24). In both of these studies, the repression was mediated by specific target sequences in the promoter region. In addition, studies of myb-dependent G 1 /S progression suggested that inhibition of the plasma Ca 2ϩ -ATPase pump by c-myb lies at the core of its ability to raise late G 1 intracellular calcium levels that in turn trigger the onset of the S phase (16).
Apart from its DNA binding effects, the ability of c-Myb to interact with other proteins via its C-terminal domain may also be an integral part of its biological activity (41). The c-Myb protein has been shown to complex, probably through the leucine zipper region, to the virally encoded transcription factor BZLF1 and activate transcription through the AP-1 site (42). Moreover, c-Myb can stimulate transcription of the hsp70 gene through heat shock elements, although c-Myb itself does not bind directly to the mentioned sequence (21).
The present study was designed to identify genes with mybdependent expression. To concentrate on the genes demonstrating direct myb-dependent regulation and to minimize the effects of protooncogene-mediated alterations in cell cycle progression, all cell lines used for differential display analysis were growth-arrested prior to RNA harvesting. Using this approach, we identified five different bands demonstrating differential expression on display gels, which was confirmed by Northern analysis using eluted band fragments as probes. Interestingly, four of five bands demonstrated decreased expression in c-myb-overexpressing cells, while only one band demonstrated increased expression in c-myb-overexpressing cells. This finding suggests that the number of genes down-regulated by c-myb expression, under the conditions of this study, may well be greater than the number of c-myb-activated genes. Thus, the primary function of c-myb may well be to contribute to the suppression of various gene products.
In this study, we identified one of the products of differential display analysis to be the mouse TSP 2 cDNA. TSP 2 belongs to a family of five related extracellular matrix proteins encoded by separate genes (43). TSPs are thought to be involved in a number of biological processes including coagulation, adhesion, cell growth, modulation of cell-cell and cell-matrix interactions, and inhibition of angiogenesis (44 -48). In particular, TSPs 1 and 2 have been shown to possess anti-angiogenic activity (27,46,49), perhaps secondary to the presence of properdin-like type 1 modules (27) or a conserved RGD sequence able to bind to the ␣ v ␤ 3 integrin (50) in both TSP 1 and TSP 2 proteins.
In order to demonstrate that TSP 2 is indeed a target for c-myb-dependent regulation, we examined its pattern of expression as a function of the cell cycle. In these experiments we found that the time of maximal suppression of TSP 2 expression closely correlated with the known time window of c-myb expression in these cells. This observation is particularly important given that expression of TSP 1 did not show a similar tightly regulated temporal pattern of expression, suggesting that it, unlike TSP 2, is not subjected to a similar c-myb-dependent regulatory mechanism.
Despite the observed reduction in expression of TSP 2 mRNA in GREMyb compared with NIH 3T3 cells, transient transfection studies with TSP 2 promoter construct demonstrated no suppression of CAT activity in myb-overexpressing cells. To further confirm the TSP 2 promoter reporter activity studies, we analyzed transcription of the native TSP 2 gene using nuclear run-on. With this approach, we observed no significant level of repressed transcription of TSP 2 mRNA in GREMyb compared with NIH 3T3 cells, further suggesting that myb-de-pendent regulation of TSP 2 expression does not occur at the transcriptional level. To address the issue of mRNA stability, we studied TSP 2 mRNA half-life in wild type and c-myboverexpressing cells. We found c-myb overexpression was associated with a 2-fold reduction in the message half-life of TSP 2 gene. Interestingly, there was no change in the TSP 1 mRNA half-life, suggesting a specific effect of c-myb expression on TSP 2 mRNA stability.
Thus, taking into account the temporal relation between c-myb and TSP 2 expression during the cell cycle and a demonstrated association of c-myb overexpression with marked reduction in TSP 2 mRNA half-life, we conclude that c-myb itself or a c-myb-regulated gene is responsible for regulation of TSP 2 gene expression in the 3T3 cells.
Recent expression studies of v-src, c-jun, and v-myc in fibroblasts (37-39) demonstrated inhibition of TSP 1 expression by these oncogenes. In this study, we found that TSP 1 mRNA A, cells harvested from a confluent culture were seeded (4 ϫ 10 6 cells/10-cm dish) in 10% FBS-containing medium. After attaching, approximately 3-4 h, the culture medium was replaced with 0.5% FBS, 0.1% BSA for 20 h, a time point when treatment with actinomycin D (25 g/ml) was initiated. Total RNA isolated at the indicated time points (5 g from control NIH 3T3 and 15 g from GREMyb cells) was subjected to Northern analysis using TSP 1 and TSP 2 probes. B, graphical representation of the stability of TSP 2 (left panel) and TSP 1 (right panel) mRNA half-life in NIH 3T3 and GREMyb cells. PhosphorImager analysis data were plotted relative to the value of TSP 2 mRNA levels immediately prior to exposure to actinomycin D (0 h). levels were down regulated in serum-starved but not serumstimulated c-myb-overexpressing cells compared with NIH 3T3 cells.
Although we have observed that c-myb expression was associated with reduction of TSP 2 mRNA half-life, the exact mechanism of this effect remains undefined. In particular, it is not clear whether c-myb is directly responsible for this effect or whether c-myb-mediated induction or suppression of another gene expression mediates the observed reduction in TSP 2 mRNA stability. Other studies (32,45) have suggested that the induction of TSP 1 and TSP 2 mRNA levels following treatment of quiescent cells with serum in the presence of cycloheximide indicates the presence of a labile ribonuclease. One possible mechanism by which expression of the full-length c-myb leads to suppression of TSP 2 expression is by transactivating expression of such ribonucleases. On the other hand, expression of the dominant negative MEn represses expression of the same molecule leading to higher levels of TSP 2 transcripts. It is interesting to note that post-transcriptional regulation plays an important role in oncogene-mediated regulation of TSP 2 expression, thus raising the need to identify such intermediate mediators of mRNA stability.
The ability of c-myb to inhibit TSP 2 expression, together with the observed effects of v-myc, c-jun, and v-src on the expression of TSP 1, may have direct implications with regard to oncogenic potential of these genes. In particular, as both TSP genes have been shown to act as suppressors of angiogenesis, inhibition of their expression may allow myb-, myc-, or srcexpressing tumors to acquire the blood supply necessary for further growth. This oncogene-mediated suppression of angiogenesis inhibitors, may therefore, play an important role in tumor growth.
Although c-myb expression is usually considered to be limited to hematopoietic cells, recent data have suggested that that is clearly not so. In fact, c-myb expression has been detected in a number of cell types including neuroretinal, smooth muscle, and colon cells (4 -6). In addition, several investigators have detected c-myb expression in fibroblasts (51,52). Although we and others (26,53) have not detected c-myb expression in NIH 3T3 cells, we have been able to show the functional presence of c-myb-like activity in these cells. In these studies, introduction of c-myb-specific dominant-negative constructs resulted in suppression of cell growth and cell cycle progression in 3T3 cells (26). Thus, we believe that there is c-myb or a c-myb-like activity in 3T3 fibroblasts.
The functional relevance of c-myb-regulated TSP 2 expression can be especially apparent in the case of vascular smooth muscle cells, a cell type demonstrating significant expression of both TSP 2 (54) and c-myb as well as a documented c-myb-dependent regulation of growth (36). In this case, c-myb-dependent suppression of TSP 2 may not only play a role in regulation of smooth muscle cell growth themselves, but also, by virtue of decreased amount of secreted TSP 2, allow growth of surrounding cells (such as endothelial cells or pericytes, cell types found in the vessel wall).